Momentum AI(M1)

DASHBOARD

What it does. Momentum is an agent that decides what you should work on next and keeps that decision current without being asked. You give it projects, objectives, deadlines and tasks; M1 - a Strands Agents SDK agent on Amazon Bedrock - ranks every open task on an Eisenhower Matrix, explains why, and names the concrete next action. A background watcher re-runs M1 for every project on a schedule and stays silent unless a deadline is genuinely at risk

then, and only then, it emails you.

Who it's for. Anyone juggling several projects with real deadlines: solo builders, students, freelancers, small teams. People who already have a task list and still spend the first 20 minutes of every day deciding where to start.

How it works. The agent is a single Strands Agent with a BedrockModel (Nova Lite) and three tools: get_project_tasks (DynamoDB), days_until (deterministic date math so the model never guesses urgency) and eisenhower_quadrant. Output is enforced with structured_output_model=M1Output, a Pydantic schema that includes a needsHumanAttention flag. The same agent is invoked two ways: on demand via FastAPI on Lambda behind API Gateway, and autonomously via an EventBridge rule that triggers a watcher Lambda. Results are cached on the project in DynamoDB; alerts go through SNS. Infrastructure is Terraform; frontend is React on Amplify.

Why it matters. Prioritization is the invisible chore under every other chore. Momentum turns it into something that happens in the background, makes the safe calls itself, and only surfaces when a human actually needs to weigh in.